Try to make field matching in fontra more obvious with newtypes. Don't reinvent chunks.

[rsheeter]

Followon to #701

[dfrg]

LGTM

[anthrotype]

i'm not entirely sure what exactly this buys us besides making the use of these strings slightly more verbose. Do AxisName or LayerName differ in behavior from each other or from a regular string?

[cmyr]

I'm with cosimo here, I'm not sure this is buying us anything.

[rsheeter]

I feel strongly that leaving it as naked String is misleading and error prone. There is at least one spot in fontra already where pythong declares str, no comment, and I had to look at actual data to see if it was a tag an axis name. I'm not especially tied to using a new type so I will update to type aliases.

[anthrotype]

There is at least one spot in fontra already where pythong declares str, no comment, and I had to look at actual data to see if it was a tag an axis name.

String is certainly not a Tag. And both a layer name and an axis name, unlike a tag, don't have any particular requirements beyond being a string. So it would be correct to type both as a String in my opinion. You can argue that the python code you were looking at was ambiguous in that it seemed to be using str type for an axis tag (which it wasn't in the end).

Also, by the way, the two type aliases are just that, alternative names to the same type, which is still String. One could still use a LayerName or an AxisName (or String) interchangeably and nothing would happen -- and that's fine.

[rsheeter]

The goal is merely it's a little bit clearer what matches what. You can still do the wrong thing but perhaps it's a little bit less likely.

For example, in python I see Location = dict[str, float] without a comment. What's the key? - maybe you know from convention it's likely to be a name but if you aren't aware of that convention it's not at all obvious. If it said Location = dict[AxisName, float] or had a comment it would be more clear. I hope using the type aliases - I agree they don't need to be newtypes - is similarly helpful. If not ... well, it probably doesn't do any harm as far as I can tell.